Software Installation and Requirements

This section outlines the requirements and steps required to install the software. Multiple method of instilation, including pip, conda, Docker and manual instillation are available, and should be selected according to the users requirements (pip and conda installs are recommended for development and Docker is recomended for using the tool). Below are the software dependencies and corresponding installation instructions for different environments.

Note

We are developing a desktop application for BRAD that can be installed without requiring the any environment or package management by the user.

Software Requirements

The software relies on the following packages and libraries:

  • Python 3.11 or higher

  • Various Python libraries (listed below)

  • pip or conda for package management

  • Docker (optional, for containerized deployment)

Dependencies

The following libraries are required and can be installed using pip or conda. These are listed in the requirements.txt and environment.yml files.

Key Libraries:

  • transformers

  • torch>=2.0.1

  • sentence-transformers

  • scikit-learn

  • matplotlib

  • seaborn

  • gget

  • biopython

  • requests

  • beautifulsoup4

  • semantic-router

  • langchain

Refer to the full list of dependencies in the requirements.txt and environment.yml files for exact versions.

Pip Installation

To install the required dependencies using pip, follow these steps:

  1. Ensure Python 3.11 or higher is installed. - You can download Python from python.org.

  2. Install dependencies from `requirements.txt`:

    First, clone the repository and navigate to the root directory:

>>> git clone <repo_url>
>>> cd BRAD
>>> pip install -r requirements.txt

This will install all necessary Python packages, including torch, transformers, langchain, and more.

Conda Instillation

Alternatively, you can install the dependencies using conda. A conda environment file is provided to easily set up the project environment:

Install Conda: If Conda is not already installed, download and install it from here.

In the root directory of the project, run the following command to create an environment named BRAD-1

>>> conda env create -f environment.yml
>>> conda list
>>> conda activate BRAD-1

Docker Instillation

>>> docker build -t brad:local .
>>> docker run -it brad:local